Comparative polyreactivity of human serum and IVIg
	ELISA reactivity
			Purified autoantibodies
Target	serum	IVIg	Polyspecific	Ferritin-specific
Human proteins
α2-acid glycoprotein	±	±	±	+++
ferritin	0	+++	+++	+++
fibronectin	0	0	±	+++
thyroglobulin	0	±	++	++
Animal proteins
actin	±	++++	++++	++++
casein	+++	++++	++++	++++
histone	±	++	+++	++++
transferrin	0	±	+++	+++
Others
dsDNA	±	+++	++++	++++
KLH	+	++++	++++	++++
LPS	+++	+++	+++	+++
Polyreactivity was evaluated by ELISA using 25 µg/mL of IgG present in IVIg and human serum. The reactivity was quantified and compared to a scale of polyreactivity established as the following: ++++, more than 50 times the background O.D.; +++, between 10 and 50 times the background O.D.; ++, between 7 and 10 times the background O.D.; +, between 5 and 7 times the background O.D.; ±, between 2 and 5 times the background O.D. and; 0, below 2 times the background O.D.

In additional experiments, we compared the relative reactivity of IVIg andhuman serum with soluble human plasma proteins (ferritin and thyroglobulin) and withother antigens (bovine casein and DNA). The results (Figure 1) are expressed as doseresponse curves for concentrations of IgG between 0.5 µg/mL and 10 mg/mL. Theresults indicated that the relative differences in reactivity between IVIg and humanserum are much more important for soluble human plasma proteins (Figure 1 A, ferritin,and 1B, thyroglobulin) than for the two other antigens (Figure 1C, casein and 1D,DNA). Indeed, IVIg were about 1000 fold more reactive for ferritin than human serum.This ratio was about 250 for thyroglobulin but only 60 for casein and about 10 forDNA. This difference between the human antigens and the others indicated that themechanisms controlling the polyreactivity of serum antibodies are much more effectiveagainst autoreactive antibodies than against polyreactive antibodies recognizingantigens not normally present in human plasma (e.g. casein and DNA). In the followingexperiments, we used purified human ferritin as an antigen model for all the other onesthat are present in human plasma.

Inhibition of IVIganti-ferritin reactivity by human serum

We determined the ability of a fixed volume of human serum (25 µg/mL ofIgG) to inhibit the reactivity of increasing amounts (10 to 500 µg/mL) of IVIg. Theobservation that the serum exhibited a very low anti-ferritin reactivity at 25 µg/mL ofIgG (Figure 1) permitted to focus on the IVIg reactivity. Results are shown on Figure 2.In panel A, the OD results indicated that the addition of the fixed amount of serum toincreasing doses of IVIg resulted in a significant reduction of the anti-ferritin reactivityat all IVIg doses tested. The shapes of the curves obtained suggested that the inhibitionwas more important at lower doses of IVIg. Indeed, the results when expressed as apercentage of inhibition by serum at each IVIg dose (Figure 2B) showed that theinhibition was high (> 75%) up to a dose of IVIg of about 50 µg/mL of IVIg was added.The inhibition then gradually decreased to reach a percentage of 40% at the maximaldose of IVIg tested (500 µg/mL). These results confirmed the ability of human serumto inhibit autoantibodies¹⁶ and further indicated that the ability of the serum to controlexogenously added IgG has limits. The saturating curve (Figure 2B) indicated that theamount of serum IgG can be increased by a factor of about 3 (25 µg/mL of endogenousIgG versus 50 µg/mL of exogenous IgG) before starting to detect a significantin vitroautoreactivity of the serum IVIg blend.

Purification of autoantibodies reacting with human serum proteins

In preliminary experiments, we observed that the small proportion of theantibodies (about 1%) reacting with bovine casein could be easily purified from IVIgusing affinity chromatography on columns of casein-Sepharose. For direct relevance tohuman patients, human serum was depleted of IgG by chromatography on protein G-Sepharose.The serum proteins were cross-linked in bulk to CNBr-activated Sepharose,which was used to purify the autoantibodies present in IVIg. Although the procedureshould purify all the autoantibodies to plasma proteins present in IVIg, we used for theassay of the fractions (total, flow-though and column eluate) the ferritin ELISA.Representative results for the anti-ferritin activity of the various fractions are shown inFigure 3. The affinity chromatography resulted in a very significant depletion of ferritinautoantibodies as shown by the shift of the flow-through curve to the right. Conversely,the ferritin autoantibodies were enriched in the eluate fraction as shown by the shift tothe left. The observation that the curves had nearly linear dose-response regions at lowOD values (< 0.4) permitted the calculation of the purification results that are listed inTable 2.

Purification of autoreactive IgG
Fractions	Total IgG (mg)	Total reactivity (U)	Specific activity (U/mg)	Purification (X)	Yield (%)
Starting IVIg	14.24	172606	12121	-	-
Flowthrough	11.30	13831	1224	0.10	8.0
Eluate	0.42	136197	327869	27.05	78.9
IVIg were chromatographied on a column of Sepharose coupled to human serum proteins. The flow through and glycine pH 2.5 eluate fractions were recovered and analyzed for their IgG content. Quantification of reactivity was estimated using ELISA and human ferritin as coating antigen. One unit of reactivity represents the amount of IgG needed to obtain an O.D. of 0.4.

As indicated by the ELISA results of Figure 3, the chromatographydepleted the IVIg of more than 90% of the anti-ferritin activity. The glycine-HCl eluatecontained about 3% of the starting IgG but more than 75% of the anti-ferritin activity.The calculated purification factor (27 X) is only indicative since it is likely much higherdue to the fact that the IgG present in the eluate are expected to react with severalplasma proteins and not only with ferritin. Thus the autoantibodies in IVIg that reactwith serum proteins can be greatly enriched by affinity chromatography.

Diversity of serum proteins recognized by autoantibodies

To evaluate the diversity of soluble auto-antigens present in serum, weperformed Western blot experiments with proteins present in IgG-depleted serum. Theserum protein blots were probed with the various fractions and the binding of antibodieswas detected with an anti-human IgG conjugate. The results are shown on Figure 4.We had to perform optimisation experiments to reduce the high background level,which was observed with the IVIg fraction. The results obtained with IVIg showed thepresence of major bands of different molecular weights along with a diffuse staining ofthe blot suggesting the presence of many minor autoantigens. The intense band of 65kilodaltons observed with all tested antibody samples corresponds to albumin and ismost likely due to non-specific binding caused by the high albumin content of theblotted proteins. The affinity chromatography was effective in depleting theautoantibodies as seen by the absence of several bands and the clearer background. Thepurified autoantibodies reacted with several autoantigens with a pattern similar to thestarting IVIg. Finally as expected from the ELISA results, serum IgG react only weaklywith some of the separated proteins giving a pattern similar to the flow-through fractionof the affinity column. This analysis showed that the autoantibodies present in IVIg andin the affinity column eluate can interact with multiple proteins in human serum incontrast to the low reactivity of human serum.

Soluble serum ICin presence of exogenous IVIg

The findings that the ferritin reactivity of IVIg was strongly inhibited by thepresence of serum (Figure 2) and that IVIg reacted with multiple plasma proteins asdetected in Western blot experiments (Figure 4) suggested the formation of IC inmixtures of serum and IVIg. This possibility was tested directly by precipitating the ICin presence of 2.5% PEG as previously reported^17,19. In preliminary experiments, weused IgG-depleted serum to ensure that the IgG present in IC originated from the addedIVIg. Although the addition of serum strongly inhibited the reactivity of IVIg, theremaining reactivity of IVIg was mostly (> 60%) found in the PEG precipitateindicating that the added IVIg could form IC with serum proteins. A similar result wasobtained with purified autoantibodies. To rule out the possibility that these results werecaused by the absence of endogenous serum IgG, the experiment was repeated withserum and biotinylated IVIg. The total and ferritin-reactive biotin-IVIg were assayedusing a streptavidin conjugate in order to detect only the added IVIg. The results (Figure5) first showed that the PEG treatment did not precipitate much IgG in the starting IVIgpreparation (< 2 %). However, almost 20 % of the IVIg added to the serum was foundin the PEG precipitate. As to the ferritin reactivity, most of the anti-ferritin IgG werefound in the PEG supernatant of IVIg. In the experimental conditions used, the additionof serum resulted in a 50 % inhibition of IVIg reactivity with ferritin. But the PEGprecipitate contained almost two times more ferritin reactivity. This result indicated theformation of soluble IC containing exogenously added IVIg in mixtures of serum andIVIg. A consistent observation in these experiments was the fact that the combinedferritin reactivity of the two PEG fractions was higher that the one of the starting serum-IVIgmixture. This result suggested that the reactivity of the soluble IC with ferritin wasincreased following their isolation by PEG treatment.

Interactions of IVIg and purified auto-IgG with complement components in presence ofhuman serum

The soluble IC formed in human serum by IVIg and purified auto-IgG couldinteract with complement components and consequently reduce the amount ofcomplement components available for pathogenic effects. The interaction withcomplement components was studied using two established assays. The Raji cell CR2assay measures the cell uptake of IC through the CR2. The results obtained (fig 6, panelA) indicated a very low percentage (<4%) of IgG-positive cells after incubation witheither serum, IVIg, purified auto-IgG or a mixture of IVIg and serum. Howeverincubation in presence of a mixture of purified auto-IgG and serum resulted in stronglyIgG positive Raji cells (about 75 %) indicating that the ability of the auto-IgG-containingIC to interact with complement components is significantly higher that theone of the IC formed with IVIg. The commercial C1q binding assay measures the relative amount of IgG complexes that can bind to immobilized C1q. The resultsobtained with the above fractions (fig. 6, panel B) showed the binding of a significantamount of IgG in presence of IVIg alone and of the mixture of IVIg and serum.However the presence of the auto-IgG fraction resulted in much more bound IgG (4-6X). The binding of an even higher amount of IgG in presence of isolated IVIg and auto-IgGfractions indicated that the polyreactive IgG present in IVIg and purified auto-IgGmay directly bind the C1q molecule. Finally it should be pointed out that the higherreactivity of the purified auto-IgG in the two assays compared to IVIg was obtained inpresence of a 40 times lower dose of IgG with the purified auto-IgG fraction (0,15mg/mL versus 6 mg/mL for IVIg).

Polyspecificity of ferritin autoantibodies

To determine if the autoantibodies reacting with various serum proteins(Figure 4) and present in IC (figure 5) are polyspecific or represent a mixture of moremonospecific antibodies, we purified the ferritin-specific autoantibodies from IVIgusing chromatography on ferritin-Sepharose. The procedure resulted in a 100-foldpurification of ferritin autoantibodies and was efficient since the flow through fractioncontained less than 5 % of the ferritin antibodies present in the starting IVIg. Thepolyspecificity of the purified anti-ferritin and anti-serum proteins was compared inELISA using the antigen panel used above. The results (Table I) showed that thepurified anti-serum proteins had a pattern of reactivity similar to the starting IVIg. Theferritin-specific autoantibodies reacted strongly with all tested structures including threeother human serum proteins (α2-acid glycoprotein, fibronectin and thyroglobulin). Thispolyspecific reactivity was confirmed in Western blot experiments in which weobserved, with anti-ferritin autoantibodies, a pattern of bands similar to the onesobtained with IVIg and purified anti-serum proteins (Figure 4). Thus, the ferritinautoantibodies are polyspecific indicating that they could form IC containing severalserum proteins.

Discussion

Our results show that the addition to human serum of doses of IVIg similarto the ones observed in the plasma of IVIg-treated patients, resulted in the formation ofIC with soluble plasma proteins. These IC were apparently formed because the addedamounts of purified IgG exceeded the ability of serum IgM to inhibit the autoreactiveIgG through id-anti-id interactions. These reactive autoantibodies could be convenientlypurified from IVIg through affinity chromatography on immobilized serum proteins orferritin. Furthermore, the soluble IC were shown inin vitro assays to interact withcomplement proteins. Additional work is necessary to better characterize the biological activity of the autoantibodies but these results permit to draw some conclusions aboutthe possible involvement of the autoIC in the modes of action of IVIg in diseasescharacterized by modulation of FcγR functions and of complement activation.

Previous work on the modulation of FcγR functions by IVIg has beenfocussed mainly on id-anti-id interactions, which could lead to the formation of IgGcomplexes^14-16. Our finding that such complexes could also be formed by interaction ofautoantibodies and soluble plasma proteins reveals an additional source of IgGcomplexes, which could have increased FcγR modulating activity. Indeed the autoIC areexpected to have a larger size and contain several IgG molecules for more efficientinteraction with FcγR. Additional structural characterization of the autoIC will permit toconfirm this hypothesis. Plasma IC have been observed in many autoimmune diseases¹⁹but the possible formation of soluble IC containing IVIg and plasma proteins has notbeen much studied so far. It is possible that these IC are rapidly cleared from circulationafter interaction with FcγR-bearing cells. However, there is evidence that autoIC may beinvolved in the therapeutic effects of IVIg in some diseases. The reactive macrophageactivation syndromes are characterized by a massive increase in plasma ferritin level (upto 10 mg/mL instead of < 1 µg/mL in healthy individuals). It was recently reported thatthe successful treatment of this disease by injection of large doses of IVIg (0,5-1 gr/kg)was related to the immune clearance of ferritin, which could be detected in plasma ICthe day after IVIg injection²¹. Our results on the inhibition of IVIg autoreactivity byserum support an involvement of autoIC in the mode of action of IVIg in other diseases.Large doses of IVIg (1-2 gr/kg) are used in the treatment of autoimmune andinflammatory diseases (reviewed in¹¹). Our results are in agreement with a previousstudy¹⁵ showing that the autoantibody inhibitory IgM present in serum must first besaturated before exogenously added IgG can form autoIC. The results (Figure 2)indicated that the autoreactivity of added IVIg is detected only after addition of a doseof IVIg containing about two times the endogenous amount of serum IgG. At this ratio,it is expected that the formation of autoIC would be optimal since further addition ofautoantibodies results in proportional increase in ELISA reactivity of the serum-IVIgblend. The observed 2:1 proportion is similar to the plasma IgG increase observed inpatients treated with about 1 gr/kg of IVIg.

Modulation of complement activation by IVIg has been shown to play a rolein the therapeutic effect of IVIg in several inflammatory diseases. The results obtainedin thein vitro assays indicate that the autoantibodies present in IVIg may play a role inthis modulation. Indeed in the Raji cell binding assay which detects the presence ofcomplement components in IgG complexes, only the mixture of purified autoantibodiesand serum was highly reactive indicating that the soluble autoIC formed may interact with complement components. The results of the C1q assay showed the strong bindingof purified autoantibodies in presence and absence of serum. It remains to be seen if thebinding in absence of serum is due to the presence of IgG complexes in purifiedautoantibodies or to the recognition of the C1q molecule by the polyreactiveautoantibodies. A consistent observation in the above assays was the increasedreactivity of the purified autoantibodies compared with proportional amounts (40 timesmore IgG) of IVIg. The reason for this difference is unclear but it could indicate that thepurification process removes some inhibitory molecules present in the IVIgpreparations. The beneficial effects of IVIg in inflammatory diseases is thought to bedependent of its ability to scavenge complement fragments such as C3b and C4b, thuspreventing their deposition in the tissue targeted by the pathogenic process². The aboveresults are consistent with this mechanism and further indicate that the autoantibodiespresent in IVIg may be involved in this process by interacting with activatedcomplement components either directly or through the formed autoIC.

The chromatography of IVIg on immobilized serum proteins yielded aneluted fraction enriched in autoantibodies, which recognized a similar diversity of serumproteins on Western blots similar as the starting IVIg. The observation that purifiedferritin autoantibodies are polyreactive and could bind to the three other serum proteinstested is significant in terms of the efficiency of the formation of IC after injection ofIVIg. It suggests that the autoantibodies can rapidly form heterogeneous IC containingvarious plasma proteins. The high diversity of recognized plasma proteins also indicatesthat the formation of IC is less likely to result in immune depletion of certain plasmaproteins in IVIg-treated patients. The purification results raise the interesting possibilityof further fractionating the current IVIg preparations into two products. The flow-throughof the column, which contains more than 95% of the starting IgG, is likely torepresent IgG reacting with non-self structures and could be used to supportimmunodeficient patients. In this regard, the monthly infusion of IVIg in those patientsis known to cause mild but significant adverse side effects in the first day followinginjection. It remains to be seen if removal of autoantibodies in IVIg could reduce theseverity of these side effects. A rare but serious adverse effect of IVIg injection isanaemia resulting from the immune destruction of the patient red blood cells caused bythe uptake of circulating IC by the complement receptor present on red blood cells²².The origin of the pathogenic IC has remained unclear. It is tempting to speculate thatthese patients may have a reduced ability to inhibit a portion of the infusedautoantibodies resulting in the formation of a higher amount of IC. The second fractionprepared by chromatography is the autoantibody eluate, which represents only about 3%of the starting IgG. This fraction could be useful in the treatment of the diseases in which IVIg have immunomodulatory roles or inhibit phagocytosis. Furthercharacterization of the biological activity of the purified autoantibodies usingin vitro(e.g. inhibition of phagocytosis²³) andin vivo (e.g. passive murine model of ITP²⁴)assays will permit to obtain the data, which could support the development of clinicaltrials in patients. These studies will reveal whether the non-autoreactive IgG present inthe flow-through fraction are important for the formation of ICin vivo. It is possible thatthese IgG contribute in the saturation of the autoantibody inhibitory mechanisms presentin serum. In this situation, the dose of autoantibodies necessary to obtain therapeuticeffects could be proportionally much larger. However, preliminary works suggest thatthis is not the case since the amount of purified ferritin autoantibodies necessary toovercome the serum inhibition was found to be about 50 times less that the amountobserved with starting IVIg (Figure 2). This observation indicated that, although theinhibitory anti-id present in serum are likely to be polyreactive, they are not able toinhibit all autoantibodies.

In conclusion, our results contribute to a better understanding of themechanisms of action of IVIg in autoimmune and inflammatory diseases and could leadto refinements in the clinical use of IVIg through preparation of IVIg sub-products fordifferent classes of diseases. This possibility would represent a significant advance inensuring the future supply of IVIg, which is currently threatened by the continuousincrease in clinical indications and market demand and by difficulties in collecting moredonor-derived plasma for production of additional IVIg.

While the invention has been described in connection with specificembodiments thereof, it will be understood that it is capable of further modificationsand this application is intended to cover any variations, uses, or adaptations of theinvention following, in general, the principles of the invention and including suchdepartures from the present disclosure as come within known or customary practicewithin the art to which the invention pertains and as may be applied to the essentialfeatures hereinbefore set forth, and as follows in the scope of the appended claims.

REFERENCES

1. Kazatchkine MD, Kaveri SV. Immunomodulation of autoimmune andinflammatory diseases with intravenous immune globulin. N Eng J Med. 2001;345:747-755

2. Larroche C, Chanseaud Y, Garcia de la Pena-Lefebvre P, Mouthon L.Mechanisms of intravenous immunoglobulin action in the treatment of autoimmunedisorders. Biodrugs. 2002;16:47-55

3. Ballow M. Intravenous immunoglobulins: clinical experience and viralsafety. J Am Pharm Assoc. 2002;42:449-459

4. Lazarus AH, Freedman J, Semple JW. Intravenous immunoglobulin and anti-Din idiopathic thrombocytopenia purpura (ITP): mechanisms of action. Transfus Sci.1998;19:289-294

5. Fehr J, Hofmann V, Kappeler U. Transient reversal of thrombocytopenia inidiotypathic thrombocytopenia purpura by high-dose intravenous gamma globulin. NEng J Med. 1982;306:1254-1258

6. Kimberly RP, Salmon JE, Bussel JB, Crow MK, Hilgartner MW. Modulationof mononuclear phagocyte function by intravenous immunoglobulin. J Immunol.1984;132:745-750

7. Samuelsson A, Towers TL, Ravetch JV. Anti-inflammatory activity of IVIgmediated through the inhibitory Fc receptor. Science. 2001;291:484-486

8. Matejtschuk P, Chidwick K, Prince A, More JE, Goldblatt E. A directcomparison of the antigen-specific antibody profiles of intravenous immunoglobulinsderived from US and UK donors plasmas. Vox Sang. 2002;83:17-22

9. Dietrich G, Algiman M, Y S, Nydegger UE, Kazatchkine MD. Origin of anti-idiotypicactivity against anti-factor VIII autoantibodies in pools of normal humanimmunoglobulin G (IVIg). Blood. 1992;79:2946-2951

10. Vassilev TL, Bineva IL, Dietrich G, Kaveri SV, Kazatchkine MD. Variableregion-connected, dimeric fraction of intravenous immunoglobulin enriched in naturalautoantibodies. J Autoimmun. 1995;8:405-413

11. Dahl MV, Bridges AG. Intravenous immune globulin: fighting antibodieswith antibodies. J Am Acad Dermatol. 2001 ;45:775-783

12. Avrameas S, Ternynck T. The natural autoantibodies system: betweenhypotheses and facts. Mol Immunol. 1993;30:1133-1142

13. Lacroix-Desmazes S, Kaveri SV, Mouthon L, Ayouba A, Evelyne M,Coutinho A, Kazatchkine MD. Self-reactive antibodies (natural autoantibodies) inhealthy individuals. J Immunol Methods. 1998;216:117-137

14. Adib M, Ragimbeau J, Avrameas S, Ternynck T. IgG autoantibody activityin normal mouse serum is controlled by IgM. J Immunol. 1990;145:3807-3813

15. Berneman A, Guilbert B, Eschrich S, Avrameas S. IgG auto- andpolyreactivities of normal human sera. Mol Immunol. 1993;30:1499-1510

16. Rossi F, Guilbert B, Tonnelle C, Ternynck T, Fumoux F, Avrameas S,Kazatchkine MD. Idiotypic interactions between normal human polyspecific IgG andnatural IgM antibodies. Eur J Immunol. 1990;20:2089-2094

17. Ohlson S, Zetterstrand K. Detection of circulating immune complexes byPEG precipitation combined with ELISA. J Immunol Methods. 1985;77:87-93

18. Lock RJ, Unsworth DJ. Measurement of immune complexes is not useful inroutine clinical practice. Ann Clin Biochem. 2000;37:253-261

19. Louzir H, Ternynck T, Gorgi Y, Ayed K, Avrameas S. Enzymeimmunoassay analysis of antibody specificities present in the circulating immunecomplexes of selected pathological sera. J Immunol Methods. 1988;114:145-153

20. Theofilopoulos AN. The Raji, conglutinin, and anti-C3 assays for thedetection of complement-fixing immune complexes. Methods Enzymol. 1981;74:511-530

21. Emmenegger U, Frey U, Reimers A, Fux C, Semela D, Cottagnoud P,Spaeth PJ, Neftel KA. Hyperferritinemia as indicator for intravenous immunoglobulintreatment in reactive macrophage activation syndromes. AmJ Hematol. 2001;68:4-10

22. Kessary-Shoham H, Levy Y, Shoenfled Y, Lorber M, Gershon H. In vivoadministration of intravenous immunoglobulin (IVig) can lead to enhanced erythrocytesequestration. J Autoimmun. 1999; 13:129-135

23. Hadley AG, Kumpel BM, Merry AH. The chemiluminescent response ofhuman monocytes to red cells sensitized with monoclonal anti-Rh(D) antibodies. ClinLab Haematol. 1988;10:377-384

24. Mizutani H, Engelman RW, Kurata Y, Ikehara S, Good RA. Developmentand characterization of monoclonal antiplatelet autoantibodies from autoimmunethrombocytopenia purpura-prone (NZW x BXSB) F1 mice. Blood. 1993;82:837-844